High resolution photos from a digital camera.

On 6 Nov 2005 10:40:23 -0800, Scott W wrote:

There will be no parallax if you rotate the camera around the nodal
point of the lens, this will be close to the front surface of the len,
normally back in a bit.

You don't want to use evenly spaced heights, that would be a cause of a
lot of parallax.

Are you're talking about *any* camera/lens, or ones that allow
tilting & shifting? I'm afraid that if I'm at ground level and
rotate *my* camera around any point, nodal or otherwise, so as to
get a picture of the top of a tall building, where will be quite a
large amount of error. We may be talking about different things,
but I thought that I was addressing PcB's concern about the ability
of commonly used cameras ("35mm or equivalent (digital or film,
doesn't matter)") to not be able to duplicate what can be done with
some large format cameras ("5x4 camera (complete with lens tilt,
etc)"). If, as I proposed, the pictures could be taken at a greater
distance, or from different heights at a close distance, there would
still be some distortion, but it would be considerably reduced,
making any further corrections in software somewhat easier.

ASAAR wrote:
On 6 Nov 2005 10:40:23 -0800, Scott W wrote:

There will be no parallax if you rotate the camera around the nodal
point of the lens, this will be close to the front surface of the len,
normally back in a bit.

You don't want to use evenly spaced heights, that would be a cause of a
lot of parallax.

Are you're talking about *any* camera/lens, or ones that allow
tilting & shifting? I'm afraid that if I'm at ground level and
rotate *my* camera around any point, nodal or otherwise, so as to
get a picture of the top of a tall building, where will be quite a
large amount of error. We may be talking about different things,
but I thought that I was addressing PcB's concern about the ability
of commonly used cameras ("35mm or equivalent (digital or film,
doesn't matter)") to not be able to duplicate what can be done with
some large format cameras ("5x4 camera (complete with lens tilt,
etc)"). If, as I proposed, the pictures could be taken at a greater
distance, or from different heights at a close distance, there would
still be some distortion, but it would be considerably reduced,
making any further corrections in software somewhat easier.

The distortion is easily taken care of by the stitching program. The
stitching programs lets you set where the effective camera is pointing,
if it points towards the horizon then there is no distortion.

I will post an example when I get a bit of time.

Scott

Scott

"Eatmorepies" writes:

If people are moving around too much they might end up in several
places in the picture.

A technique that produces interesting pictures. Stand on the beach
and take photos of a child running into the sea, pan the
camera. Stich them together and get the child in 4 or 5 different
places.

If that's the effect you want, then sure. If you don't want it,
better be careful. You could also use the technique to entirely
remove moving objects from a photo.

--
Måns Rullgård

Scott W wrote:
Yesterday I took a 95 MP photo using my digital camera, here is a link
to a overview photo along with a small 100% crop from the photo.
http://www.pbase.com/konascott/image/51841148/original

And only three $^%# cell phones in sight.

The photos is 15730 by 6000 pixels, just short of a 100 MP photo, it is
a view of the small beach in front of the King Kamehameha hotel, taken
off the Kailua Pier in Kona Hawaii.

One can stitch together images from a 4x5 camera as well.

--
Ray Fischer

On 6 Nov 2005 11:37:51 -0800, Scott W wrote:

The distortion is easily taken care of by the stitching program. The
stitching programs lets you set where the effective camera is pointing,
if it points towards the horizon then there is no distortion.

I will post an example when I get a bit of time.

Ok, but take it easy. My puny modem doesn't take kindly to being
abused.

In article ,
ASAAR wrote:
On 6 Nov 2005 10:40:23 -0800, Scott W wrote:

There will be no parallax if you rotate the camera around the nodal
point of the lens, this will be close to the front surface of the len,
normally back in a bit.

You don't want to use evenly spaced heights, that would be a cause of a
lot of parallax.

Are you're talking about *any* camera/lens, or ones that allow
tilting & shifting? I'm afraid that if I'm at ground level and
rotate *my* camera around any point, nodal or otherwise, so as to
get a picture of the top of a tall building, where will be quite a
large amount of error.

Panotools takes care of that. It assembles the images onto the inside of a
sphere and resamples them according to your viewpoint.

On Sun, 06 Nov 2005 22:27:14 GMT, Chris Brown wrote:

Panotools takes care of that. It assembles the images onto the inside of a
sphere and resamples them according to your viewpoint.

Yes, I'm aware that software is available that can make those
adjustments. But there will be much less detail available at the
uppermost part of the frame. That is, if the distance is such that
people are just barely recognizeable if they're located near windows
on the ground floor, people similarly positioned near windows on the
uppermost floor won't be. Might not even be recognizeable as
people. That's probably satisfactory for many stitched shots, but
the two methods I mentioned would result in more uniformly high
resolution from top to bottom. If it's possible to back up
sufficiently and use a longer lens, there's very little additional
effort required. If not, the other method (taking shots on
successively higher floors) might be too much effort, and might not
even be allowed without special permission.

ASAAR wrote:
On Sun, 06 Nov 2005 22:27:14 GMT, Chris Brown wrote:

Panotools takes care of that. It assembles the images onto the inside of a
sphere and resamples them according to your viewpoint.

Yes, I'm aware that software is available that can make those
adjustments. But there will be much less detail available at the
uppermost part of the frame. That is, if the distance is such that
people are just barely recognizeable if they're located near windows
on the ground floor, people similarly positioned near windows on the
uppermost floor won't be. Might not even be recognizeable as
people. That's probably satisfactory for many stitched shots, but
the two methods I mentioned would result in more uniformly high
resolution from top to bottom. If it's possible to back up
sufficiently and use a longer lens, there's very little additional
effort required. If not, the other method (taking shots on
successively higher floors) might be too much effort, and might not
even be allowed without special permission.

For those who stitch pixels are cheap, just use a longer lens and
stitch more photos.

Scott

Måns Rullgård wrote:
"Eatmorepies" writes:

If people are moving around too much they might end up in several
places in the picture.

A technique that produces interesting pictures. Stand on the beach
and take photos of a child running into the sea, pan the
camera. Stich them together and get the child in 4 or 5 different
places.

If that's the effect you want, then sure. If you don't want it,
better be careful. You could also use the technique to entirely
remove moving objects from a photo.
One of the things I want to try sometime it to take a lot of photos of
a very busy road
and by combining the right photos together remove all the cars but
leave the people on
the sidewalks, I think it might make for an interesting photo.

Scott

ASAAR wrote:
On 6 Nov 2005 11:37:51 -0800, Scott W wrote:

The distortion is easily taken care of by the stitching program. The
stitching programs lets you set where the effective camera is pointing,
if it points towards the horizon then there is no distortion.

I will post an example when I get a bit of time.

Ok, but take it easy. My puny modem doesn't take kindly to being
abused.

I kept these small.
Both of these photos are stitched from the same 4 photos, I could not
get a wide enough angle view with just one photo and the lens I was
using.

In this first view I have put the view point looking halfway up the
building, like it would be if the shot were taken with normal camera.
http://www.pbase.com/konascott/image/51887904/original

In this next view I have put the view point looking at about the door.
http://www.pbase.com/konascott/image/51887863/original

There are real limits as to how much of this you can do before it
starts to look odd, either with a view camera or stitching software.
In fact the photos from both will look the same, really.

Scott